Hyperglycemia Activates p53 and p53-Regulated Genes Leading to Myocyte Cell Death

Left ventricular myocytes were isolated from male Sprague-Dawley rats at 3 months. Under chloral hydrate anesthesia (300 mg/kg body wt), hearts were excised, and myocytes were dissociated by collagenase (11,24). Trypan-blue–excluding cells constituted 85%. Nonmyocytes accounted for 1–2%. Myocytes were cultured in serum-free medium (SFM) and in the absence of insulin. After 30 min, the medium was changed with SFM containing 5.5, 12.5, or 25 mmol/l glucose, corresponding to plasma levels of 100, 225, and 450 mg/dl, respectively. Hyperosmolarity was corrected by decreasing salts, and acidosis was corrected by adding 20 mmol/l HEPES. Conversely, hyperosmolarity was mimicked by including 20 mmol/l mannitol. The effect of the AT₁ antagonist losartan (10⁻⁷ mol/l), O-glycosylation blocker, benzyl 2-acetamido-2-deoxy-α-d-galactopyranoside (BAG, 4 mmol/l; Sigma, St. Louis, MO), and p38-MAP kinase (MAPK) inhibitor (SB 202190, 1 μmol/l; Calbiochem, San Diego, CA) were determined; these compounds were added to the medium 1 h before glucose. Cultures were studied at 10, 20, and 45 min and at 1, 3, 12, 24, 48, and 96 h. Myocytes were fixed in 1% formaldehyde for cytochemistry, and they were frozen at −80°C for biochemistry (11,12,22,24).

Myocytes were suspended in lysis buffer consisting of 50 mmol/l Tris-HCl, pH 7.5, 5 mmol/l EDTA, 250 mmol/l NaCl, 0.1% Triton X-100, 2 mmol/l phenylmethylsulfonyl fluoride (PMSF), 2 μg/ml aprotinin, 5 mmol/l dithiothreitol, and 1 mmol/l Na₃VO₄. A volume of 50 μg proteins were separated on 12% SDS-PAGE and transferred onto nitrocellulose. Carbohydrate moieties were oxidized with 10 mmol/l sodium periodate, followed by incorporation of biotin hydrazide. Biotinylated complexes were detected by streptavidine-alkaline phosphatase and BCIP/NBT color development reagents (Bio-Rad, Hercules, CA). Protein glycosylation was confirmed by performing enzymatic deglycosylation with O-glycosidase D and hexosaminidase, before periodate oxidation (25). Glycosylated p53 was detected by short-term labeling of adherent myocytes with [³⁵S]methionine followed by immunoprecipitation of cell lysates with a p53 monoclonal antibody, 3 μg pAb 240 (Santa Cruz Biotechnology, Santa Cruz, CA). Glycosylated residues bound to p53 were recognized by immunoprecipitation of proteins with 3 μg pAb 240; Western blot was done with the monoclonal antibody RL2 anti–O-linked GlcNAc (Alexis, San Diego, CA) (26).

A volume of 50 μg proteins were separated on 12% SDS-PAGE, transferred to nitrocellulose, and exposed to three distinct phosphorylated p53 antibodies with 1 μg/ml in Tris-buffered saline/Tween 20 (TBST). Rabbit phosphorylated p53 antibody (New England BioLabs, Beverly, MA) recognizes phosphorylation of p53 at Ser 390 only (15). This site, Ser 390, in rodents corresponds to Ser 392 in humans. This polyclonal antibody is produced by immunizing rabbits with a synthetic phosphorylated Ser 392 peptide corresponding to residues around Ser 392 of human p53. Mouse p53 antibody 421 (Calbiochem) is a monoclonal antibody generated by immunizing mice with p53 protein. The epitope against which the antibody is raised is located at the COOH terminus of p53, from amino acids 372 to 381 (27,28). This region overlaps with the phosphorylation site of protein kinase C (PKC) in the regulatory domain of p53. Thus, antibody 421 does not react with p53 after phosphorylation by PKC (9). Rabbit phosphorylated p53 Ser 15 antibody (Ab-3; Oncogene, Boston, MA) is generated by immunizing rabbits with a synthetic phosphorylated Ser 15 peptide corresponding to residues surrounding Ser 15 of human p53. Human Ser 15 corresponds to Ser 18 in rodents. Phosphorylated p53 Ser 15 antibody detects p53 only when phosphorylated at Ser 15 and does not react with nonphosphorylated p53 (17). Bound antibodies were detected by horseradish peroxidase (HRP)-conjugated anti-mouse or anti-rabbit IgG. Phosphorylated p53 and nonactivated p53 were detected as 53-kDa bands.

Myocyte lysates (80 μg), were exposed to rabbit phosphorylated p38-MAPK antibody (New England BioLabs) or anti-phosphorylated MAPK and anti-phosphorylated JNK polyclonal antibodies (Promega, Madison, WI) to identify the active forms of p38-MAPK, ERK1 and -2, and JNK1 and -2, respectively. Phosphorylated p38-MAPK was seen as a band at 38 kDa. Active ERK1 and -2 were not detected, whereas active JNK1 and -2 were recognized as two close bands at nearly 50 kDa. To evaluate the enzymatic activity of p38-MAPK, protein lysates, 60 μg, were exposed to immobilized phosphorylated p38-MAPK antibody. Pellets were resuspended in kinase buffer containing 200 μmol/l ATP and 2 μg activating transcription factor-2 (ATF-2) and were kept at 30°C for 30 min. Samples were separated on 12% SDS-PAGE, transferred onto nitrocellulose, and exposed to phosphorylated ATF-2 antibody (15).

To prepare a double-stranded probe for Bax, oligonucleotides 5′-AGCTTGCTCACAAGTTAGAGACAAGCCTGGGCGTGGCTATATTGA-3′ and 5′-AGCTTCA-ATATAGCCCACGCCCAGGCTTGTCTCTAACTTGTGAGCA-3′, were annealed and labeled with [γ-³²P]ATP and T4 poly-nucleotide kinase. Oligonucleotides 5′-GCTGAGCTTGGATCTGG-AAGGCGACACTGGG-3′ and 5′-CCCAGTGTCGCCTTCCAGATCCAAGCTCAGC-3′ were used to prepare a AT₁ probe. Oligonucleotides 5′-AGCCTCTGTACAGAGTAGCCTGGGAATA-GATCCATCTTC-3′ and 5′-GAAGATGGATCTATTCCCAGGCTACTCTGTACAGAGGCT-3′ were used for Aogen probe (11,22). Nuclear extracts were incubated with [³²P]-labeled probes and were separated on 4% PAGE. For specificity, nuclear extracts were exposed to a p53 antibody, 0.5 μg pAb 240 (Santa Cruz Biotechnology), or to an irrelevant antibody, mouse monoclonal anti-human Mdm2 (SMP14; Santa Cruz Biotechnology), and subject to competition with an excess of unlabeled self-oligonucleotide (11,22).

Proteins (50 μg) were separated on SDS-PAGE, transferred onto nitrocellulose, and exposed to mouse monoclonal p53 (pAb 240; Santa Cruz Biotechnology), rabbit polyclonal Bcl-2 (ΔC21; Santa Cruz Biotechnology), rabbit polyclonal Bax (P19; Santa Cruz Biotechnology), mouse monoclonal Aogen (Swant, Bellinzona, Switzerland), rabbit polyclonal AT₁ (306; Santa Cruz Biotechnology), goat polyclonal AT₂ (C-18; Santa Cruz Biotechnology) antibodies, and 1 μg/ml TBST. Bound antibodies were detected by HRP-conjugated anti-mouse, anti-rabbit, or anti-goat IgG. p53 was detected as a 53-kDa band, Bax as a 19-kDa band, Bcl-2 as a 25-kDa band, Aogen as a 56- to 58-kDa band, AT₁ as a 41-kDa band, and AT₂ as a 42- to 44-kDa band (11,12,22).

Ang II concentration in conditioned medium (CM) was measured by enzyme-linked immunosorbent assay (ELISA) (Peninsula, Belmont, CA). CM was treated with 10% trifluoroacetic acid (TFA) and centrifuged. The supernatant was dried, and the residue was dissolved in 0.1% TFA and purified in C-18 Sep-Pak column. Ang II fraction was eluted with 30% acetonitrile in 5 ml of 0.1% TFA, dried, and dissolved in TBST. Samples (50 μl) were analyzed in a microtiter plate using Ang II antibody (1:32,000 dilution) and a tracer, biotinylated Ang II. Color reaction was developed with tetrametyl-benzidine. Absorbance was recorded at 450 nm, and the concentration was calculated from standard curves (12,22).

Cells were incubated with 50 μl cacodylate buffer containing 5 units terminal deoxynucleotidyl transferase (TdT) and biotinylated dUTP. dUTP was detected by fluroescein isothiocyanate–avidin. Cytoplasm was stained with α-sarcomeric actin and nuclei with propidium iodide (PI). In each condition, 2000–3000 cells were examined by confocal microscopy (11,29).

An oligonucleotide probe with single-base 3′ overhang (11,22,29) was used to detect double-strand DNA breaks with single-base 3′ overhangs. Cells were incubated in 10 mmol/l sodium citrate and were treated with a solution containing 1 unit/μl T4 ligase and 35 ng/μl hairpin probe (Synthetic Genetics, San Diego, CA).

Results are the means ± SD. The significance between two measurements was determined by Student’s t test and by multiple comparisons by ANOVA and Bonferroni method. P values of <0.05 were considered significant. Unless indicated in the text or legends, n equaled five separate cell isolations in each case.

To evaluate whether p53 became glycosylated, myocytes were cultured in SFM at 25 mmol/l glucose; 5.5 mmol/l glucose was used as a control. With high glucose, osmolarity was adjusted to 290 mOsm. Glycoprotein detection by periodate oxidation showed glycosylation of a 55.6-kDa protein (Figs. 1A and B). The increase in protein glycosylation seen at 10 min with high glucose was not statistically significant. However, a threefold increase in optical density at 20 min (0.50 ± 0.20 vs. 1.7 ± 0.5, P < 0.001) and a sixfold increase at 45 min (0.42 ± 0.20 vs. 2.6 ± 0.8, P < 0.001) was demonstrated; glycosylation peaked sevenfold (0.44 ± 0.12 vs. 3.1 ± 0.7, P < 0.001) at 1 h and remained elevated fourfold (0.56 ± 0.29 vs. 2.2 ± 0.5, P < 0.001) at 3 h. A progressive decrease in glycosylation of the 55.6-kDa protein was noted at 12, 24, 48, and 96 h. Enzymatic deglycosylation before periodate oxidation prevented the appearance of the 55.6-kDa glycoprotein (Figs. 1A and B).

In the presence of high glucose, losartan did not interfere with protein glycosylation at 45 min and 1 h (Fig. 1C). Conversely, inhibition of O-glycosylation by BAG markedly attenuated the 55.6-kDa protein (Fig. 1D). The 55.6 kDa glycoprotein was identified as p53 by radiolabeling myocytes exposed to 5.5 and 25 mmol/l glucose with [³⁵S]methionine for 1, 2, and 3 h. Myocyte lysates were then immunoprecipitated with pAb 240 anti-p53 (Fig. 1E). At high glucose, two proteins at 55.6 and 53 kDa were detected. The lower–molecular weight band corresponded to p53, and the higher molecular weight band reflected the addition of carbohydrates to the p53 protein.

To identify the sugar moiety bound to p53, myocyte lysates obtained from cells cultured in the presence of 5.5 and 25 mmol/l glucose were immunoprecipitated with pAb 240 anti-p53. Immunoprecipitated proteins were subjected to Western blot with the monoclonal antibody RL2, which recognized O-GlcNAc linked to p53 (Fig. 1F). The O-glycosylated form of p53 was apparent at 55.6 kDa. O-GlcNAc-p53 was detected in myocytes exposed to high glucose for 20 min (0.8 ± 0.2, n = 5). O-GlcNAc-p53 increased at 45 min (2.2 ± 0.3, n = 5; P < 0.001), peaking at 1 h (4 ± 0.57, n = 5; P < 0.001). The level of O-GlcNAc-p53 remained stable at 3 h (3.5 ± 0.7, n = 5; P < 0.001) and decreased progressively at 12 h (1.6 ± 0.4, n = 5; P < 0.001) and 24 h (1.2 ± 0.3, n = 5; P < 0.001).

To establish whether O-glycosylation of p53 upregulated the myocyte RAS, Ang II concentration, expressed as picograms per hour per 10⁶ cells, was measured by ELISA in CM collected from myocytes cultured at 5.5 and 25 mmol/l glucose under isosmotic conditions. In comparison with 5.5 mmol/l glucose, Ang II formation at 25 mmol/l glucose did not increase at 10 and 20 min (values were combined), but increased 1.4-fold at 45 min, 1.6-fold at 1 and 3 h (values were combined), and 1.8-fold at 12 h (Fig. 2). Inhibition of O-glycosylation by BAG prevented the increase in Ang II at each time point.

The next question concerned whether p53 glycosylation–mediated Ang II formation resulted in p53 phosphorylation. Phosphorylation of p53 at the COOH-terminal is implicated in the upregulation of the myocyte RAS and the proapoptotic protein Bax as well as downregulation of the antiapoptotic protein Bcl-2 (10,11,12,13,22). This may initiate cell death (11,22). For this purpose, cell lysates, obtained from myocytes cultured at 5.5 and 25 mmol/l glucose, were exposed to p53 antibodies recognizing, respectively, the site where phosphorylation by PKC at Ser 376 occurred and the site where phosphorylation by p38-MAPK at Ser 390 occurred. p53 phosphorylation at Ser 390 was undetectable at 5.5 mmol/l glucose, nor was it detectable at 25 mmol/l glucose after 10, 20, and 45 min (Figs. 3A and B). With high glucose, p53 phosphorylation was clearly visible at 1 h (2.4 ± 0.7, n = 5). A progressive increase in phosphorylated p53 at Ser 390 was noted at 3 h (3.6 ± 0.9, n = 5; P < 0.05), 12 h (4.5 ± 1.1, n = 5; P < 0.01), 24 h (6.4 ± 1.8, n = 5; P < 0.005), 48 h (11 ± 3, n = 5; P < 0.001), and 96 h (15 ± 4, n = 5; P < 0.001). Losartan and BAG prevented phosphorylation (Figs. 3C and D).

At 25 mmol/l glucose, there was an increase with time, from 1 h to 4 days, in the level of nonactivated p53 at Ser 376 that corresponds to the site of p53 activation by PKC (Fig. 3E). The increase in the latent form of p53 with high glucose followed the decrease in p53 glycosylation, because pAb 421 anti-p53 does not recognize glycosylated p53. Glycosylated p53 peaked at 1 h, when nonactivated p53 at Ser 376 was not detectable. As expected, the amount of nonactivated p53 at 5.5 mmol/l glucose was similar to that at 25 mmol/l, reflecting the lack of p53 phosphorylation at Ser 376 with low and high glucose concentrations. Losartan did not modify the level of nonactivated p53 at Ser 376 (Fig. 3F), indicating that translocation of PKC by Ang II was not involved in p53 phosphorylation. Optical density values for nonactivated p53 are not shown.

Phosphorylated p53 at Ser 18 represents an additional site of activation of the tumor suppressor in response to DNA damage (16,17,18). Thus, phosphorylated p53 at Ser 18 was evaluated. Phosphorylated p53 was not detectable at 5.5 mmol/l glucose. At 25 mmol/l glucose, phosphorylated p53 at Ser 18 was not apparent at 45 min, but its quantity increased progressively at 1 h (0.11 ± 0.02, n = 5), 3 h (0.37 ± 0.09, n = 5; P < 0.05), 12 h (0.52 ± 0.11, n = 5; P < 0.001), 24 h (0.69 ± 0.12, n = 5; P < 0.001), 48 h (0.91 ± 0.13, n = 5; P < 0.001), and 96 h (1.36 ± 0.25, n = 5; P < 0.001) (Fig. 3G).

Phosphorylation of p53 at the COOH-terminal and at Ser 390 is mediated by the activation of p38-MAPK (15). To strengthen the notion that p53 phosphorylation by this MAPK followed p53 glycosylation–induced Ang II formation, myocytes were cultured at 5.5 and 25 mmol/l glucose in the presence of an inhibitor of p38-MAPK (SB 202190) or an AT₁ antagonist (losartan). AT₁ blockade prevents activation of p38-MAPK by Ang II. Ang II concentration was then measured under two conditions of suppressed p38-MAPK. Values at 10 and 20 min and at 1 and 3 h were comparable and, thus, combined. When phosphorylation was blocked, the production of Ang II was not decreased for up to 3 h but was inhibited at 12 h (Fig. 3H). At 12 h, p53 glycosylation was significantly decreased (Figs. 1A–C), whereas p53 phosphorylation at Ser 390 was upregulated (Figs. 3A–C). Losartan produced effects similar to SB 202190. The attenuated synthesis of Ang II at 12 h by SB 202190 and losartan documented that p53 phosphorylation was a secondary event triggered by p53 glycosylation.

To confirm that p38-MAPK was stimulated by Ang II and was responsible for p53 phosphorylation at Ser 390, phosphorylated p38 was measured by Western blot in myocytes cultured at 5.5 and 25 mmol/l glucose in isosmotic media (Figs. 4A and B). Phosphorylated p38 was undetectable at all time points with 5.5 mmol/l glucose; this was also the case at 25 mmol/l glucose after 10, 20, and 45 min. At 25 mmol/l glucose, phosphorylated p38 became apparent at 1 h (0.5 ± 0.2, n = 5) and 3 h (0.7 ± 0.3, n = 5). Phosphorylated p38 progressively increased at 12 h (1.0 ± 0.3, n = 5; P < 0.005), 24 h (1.5 ± 0.5, n = 5; P < 0.003), 48 h (2.1 ± 0.7, n = 5; P < 0.002), and 96 h (2.8 ± 0.9, n = 5; P < 0.001). p38-MAPK enzymatic activity was also assessed by measuring ATF-2 phosphorylation. ATF-2 is a physiological substrate of p38-MAPK. Phosphorylated ATF-2 was not visible in myocytes at 5.5 mmol/l glucose. However, in cells treated with 25 mmol/l glucose, activated ATF-2 increased from 1 to 96 h (Fig. 4C; optical density values not shown). Earlier intervals were not examined because phosphorylated p38 was not present. We thus established that activation of p38-MAPK was caused by Ang II synthesis induced by p53 glycosylation. Western blots were performed with myocytes treated with the inhibitor of O-glycosylation, BAG, 1 h before the exposure of the cells to low and high glucose concentrations. The active form of the enzyme was undetectable in these samples (Fig. 4D).

To complete the analysis of the effects of glucose on the MAPK family of proteins, the active forms of ERK and JNK were examined. Phosphorylated ERK1 and -2 levels were not detected in myocytes cultured in the presence of low and high glucose (data not shown). In contrast, the active forms of JNK1 and JNK2 were recognized at high glucose (Fig. 4E). This was apparent at 20 min (0.61 ± 0.08, n = 5) and consistently increased at 1 h (1.1 ± 0.11, n = 5; P < 0.05), 3 h (1.27 ± 0.22, n = 5; P < 0.01), 12 h (2.38 ± 0.31, n = 5; P < 0.001), 24 h (2.67 ± 0.34, n = 5; P < 0.001), and 48 h (2.9 ± 0.4, n = 5; P < 0.001). Phosphorylated JNK was not detected at 5.5 mmol/l glucose.

To document that p53 phosphorylation enhanced p53 function, gel mobility assays were performed. In comparison with myocytes at 5.5 mmol/l glucose, the optical density of the p53-shifted band for the Aogen promoter increased 5-fold (P < 0.001) and 12-fold (P < 0.001) in cells exposed to 25 mmol/l glucose for 1 day (0.42 ± 0.13 vs. 2.2 ± 0.6) and 2 days (0.45 ± 0.21 vs. 5.3 ± 1.9), respectively (Fig. 5A). Osmolarity did not affect p53 binding. Controls included preincubation of nuclear extracts with a p53 antibody or with an excess of unlabeled self-oligonucleotide; both conditions prevented the formation of a p53-shifted complex. Conversely, the addition of an irrelevant antibody did not interfere with the p53 band.

With respect to cells at 5.5 mmol/l glucose, myocytes cultured at 25 mmol/l glucose showed a fourfold (P < 0.001) and an eightfold (P < 0.001) increase in the p53-shifted complex for the AT₁ promoter at one (0.13 ± 0.05 vs. 0.53 ± 0.18) and two (0.18 ± 0.10 vs. 1.4 ± 0.33) d, respectively (Fig. 5B). Similarly, high glucose increased p53 DNA binding to the Bax promoter fourfold (P < 0.001) and ninefold (P < 0.001) at one (1.1 ± 0.4 vs. 4.2 ± 1.2) and two (1.3 ± 0.9 vs. 12 ± 4) d (Fig. 5C). Consistency in protein loading, lack of protein degradation, and uniformity in the relative purity of nuclear extracts were evaluated by SDS-PAGE and Coomassie blue staining (not shown).

The expression of p53, Bax, Bcl-2, Aogen, AT₁, and AT₂ receptors was evaluated in myocytes exposed to 5.5, 12.5, and 25 mmol/l glucose for 4 days. We have previously shown that conditions characterized by hyperglycemia-induced Ang II formation (30,31,32), or enhanced Ang II synthesis alone, independent from hyperglycemia (11,12,22), led to upregulation of p53 and p53-associated genes. Currently, p53 protein increased in proportion to glucose concentration (0.23 ± 0.07, 1.0 ± 0.22, and 2.2 ± 0.6 for 5.5, 12.5, and 25 mmol/l, respectively, n = 6 in all cases; P < 0.05–0.001 vs. 5.5 mmol/l) and losartan, at 10⁻⁷ mol/l, completely abolished the effect of hyperglycemia on p53 quantity (data not shown). p53 activation was paralleled by comparable increases in Bax (2.3 ± 0.5, 7.9 ± 1.7, and 13 ± 2.3 for 5.5, 12.5, and 25 mmol/l, respectively, n = 6 in all cases; P < 0.001 vs. 5.5 mmol/l), whereas an inverse correlation was observed between glucose and Bcl-2 expression (9.3 ± 1.6, 6.6 ± 1.0, and 3.7 ± 0.7 for 5.5, 12.5, and 25 mmol/l, respectively, n = 6 in all cases; P < 0.005–0.001 vs. 5.5 mmol/l); losartan prevented the changes in Bax and Bcl-2. Aogen (0.73 ± 0.18, 2.4 ± 0.4, and 3.8 ± 0.7 for 5.5, 12.5, and 25 mmol/l, respectively, n = 6 in all cases; P < 0.001 vs. 5.5 mmol/l) and AT₁ receptors (1.7 ± 0.3, 4.0 ± 0.8, and 6.0 ± 0.7 for 5.5, 12.5, and 25 mmol/l, respectively, n = 6 in all cases; P < 0.001 vs. 5.5 mmol/l) mimicked the adaptations of p53, showing higher levels with higher glucose concentrations. As expected, AT₂ receptors were not modified by glucose (3.7 ± 0.5, 4.0 ± 0.7, and 3.6 ± 0.5 for 5.5, 12.5, and 25 mmol/l, respectively, n = 6 in all cases; NS) or losartan. Hyperosmolarity did not significantly change any of the listed values.

To demonstrate that enhanced p53 function and upregulation of Aogen with hyperglycemia resulted in a long-term increase in Ang II synthesis, Ang II was measured in CM after exposure of myocytes to 5.5, 12.5, and 25 mmol/l glucose for 1, 2, and 4 days (Fig. 6). Osmolarity did not affect Ang II levels, and values were combined. In comparison with 5.5 mmol/l glucose, 25 mmol/l glucose increased Ang II formation 1.9-fold at 1 day. Additionally, Ang II generation increased 1.9- and 2.5-fold at 2 days and 2.6- and 3.9-fold at 4 days with 12.5 and 25 mmol/l glucose, respectively.

This form of cell death is characterized first by nuclear damage, which is frequently not detectable morphologically. However, double DNA strand breaks can be identified histochemically. This initial phase of the apoptotic process is followed by nuclear fragmentation and later by cytoplasmic alterations (33). To determine whether Ang II activated cell death, myocyte apoptosis was measured. Figure 7A–I illustrates by confocal microscopy three myocytes undergoing apoptosis. Nuclear morphology was preserved in Figs. 7A–C, whereas nuclear fragmentation was apparent in Figs. 7D–I. Cell striation was evident in the first example. Loss of cytoplasmic structure and cell shrinkage were seen in the other two cases.

Apoptosis was measured in myocytes cultured at 5.5, 12.5, and 25 mmol/l glucose for periods of 1, 2, and 4 days. Cell death increased with time and glucose concentration (Fig. 7J). Hyperosmolarity of the medium did not alter the degree of apoptosis at any time point or glucose concentration, including 4 days and 25 mmol/l glucose. Values collected under isosmotic and hyperosmotic conditions were combined. The lack of impact of increased osmolarity on apoptosis was confirmed by measuring cell death at 5.5 mmol/l glucose in the presence of 20 mmol/l mannitol for 4 days. Osmolarity increased to a value comparable with 25 mmol/l glucose, but apoptosis was essentially identical to that found at 5.5 mmol/l glucose, with 2.8 ± 1.2 and 3.6 ± 1.7% for control and mannitol, respectively (n = 6; P = 0.6). At 4 days of hyperglycemia, apoptosis comprised nearly 3, 8, and 15% of myocytes at 5.5, 12.5, and 25 mmol/l glucose, respectively. Moreover, to evaluate the impact of BAG on myocyte apoptosis, 4 mmol/l of this inhibitor of O-glycosylation was added to myocytes cultured at 25 mmol/l glucose in isosmotic medium. Myocyte apoptosis was measured 4 days later, with 2.5 ± 1.1% for 5.5 mmol/l glucose (n = 6), 13 ± 4% for 25 mmol/l glucose and no BAG (n = 6; P < 0.001), and 3.5 ± 1.7% for 25 mmol/l glucose plus BAG (n = 6; P = 0.7). Thus, hyperosmolarity had no effect on the magnitude of myocyte apoptosis, and BAG blocked high glucose–mediated cell death.

To examine the effects of AT₁ blockade on cell death, losartan was added 1 h before glucose and maintained for up to 4 days. Losartan attenuated apoptosis by nearly 60% in all conditions but failed to decrease cell death to baseline (Fig. 7K). These values of apoptosis were obtained by TdT. Enhanced uptake of glucose by myocytes leads to the generation of lactic acid, thereby decreasing extracellular pH (34). The effects of glucose concentration and time on the pH of the medium are shown in Fig. 7L. Essentially, 5.5 mmol/l glucose did not decrease pH during the 4-day period. However, 25 mmol/l glucose reduced pH throughout the duration of the culture. To prevent the development of acidosis, 20 mmol/l HEPES were added to the medium (Fig. 7L). The influence of pH on hyperglycemia-induced apoptosis was determined. The impact of losartan on cell death was also re-examined under low and physiological pH. TdT detects double-strand DNA cleavage induced by Ca²⁺-dependent DNase I and pH-dependent DNase II (11,29), but it cannot separate the effects of these two DNases. Conversely, the hairpin probe identifies double-strand DNA cleavage with a single-base 3′ overhang, which is the form of apoptosis mediated by Ang II exclusively via the stimulation of Ca²⁺-dependent DNase I (11,29).

Figure 7M illustrates the extent of myocyte apoptosis in cell cultures at 5.5 and 25 mmol/l glucose in the presence and absence of losartan for 4 days. These experiments were performed in isosmotic media. At 25 mmol/l glucose and pH 6.9, myocyte apoptosis measured by the hairpin probe was significantly lower than that assessed by TdT. As expected, losartan did not abolish cell death in the presence of acidosis, resulting in a significant TdT labeling of myocyte nuclei. In contrast, at 25 mmol/l glucose and pH 7.4, apoptosis evaluated by hairpin probe was almost identical to that determined by TdT. As expected, losartan reduced cell death to control values when measured by either TdT or hairpin probe. Thus, losartan completely inhibited myocyte apoptosis triggered by Ang II formation with hyperglycemia.

In view of the presence of acidosis with hyperglycemia, we evaluated the potential effects of low pH on the main pathways implicated in the activation of p53 and p53-mediated apoptosis in this model system. The levels of phosphorylated p53 at Ser 390 and phosphorylated p38 MAPK were not affected by pH in myocytes cultured for 2 and 4 days at 25 mmol/l glucose. The presence or absence of HEPES did not change the activity of these two proteins (Figs. 8A and B). These observations confirmed that the biochemical results obtained with high glucose concentration were not influenced by pH.

The findings of the present study indicate that high glucose concentrations led to O-glycosylation of p53 and synthesis of Ang II in adult cardiac myocytes. Binding to AT₁ receptors promoted the activation of p38-MAPK, which in turn induced p53 phosphorylation at Ser 390 in the COOH terminus. This post-translational modification sustained p53 function and the upregulation of p53-dependent and -regulated genes. p53 stability and transcriptional activity were also influenced by its phosphorylation at Ser 18 in the NH₂ terminus. Conversely, hyperglycemia induced a dual phosphorylation of JNK at Thr 183 and Tyr 185 (19,21), most likely facilitating the degradation of p53 (35,36). However, these multiple responses enhanced p53 activity, increasing the susceptibility of myocytes to undergo Ang II–mediated apoptosis. The extent of myocyte death directly correlated with glucose and Ang II levels. With hyperglycemia, intracellular acidosis became apparent, and this metabolic defect accounted for nearly 40% of the myocyte death. The AT₁ blocker losartan prevented the effects of Ang II on cell death but had no influence on apoptosis triggered by acidosis. Importantly, inhibition of O-glycosylation counteracted this cascade of events by interfering with the initiation of p53 function and the generation of Ang II. Thus, hyperglycemia upregulated the myocyte RAS, which is implicated in the development of a diabetic cardiomyopathy in vivo (30,31,32).

Glycosylation of proteins with hyperglycemia in vivo and in vitro occurs through the addition of carbohydrates, without involving an enzymatic reaction (1,2). This process is commonly defined as glycation or nonenzymatic glycosylation. However, the formation of glycation products requires time and is not evident before cells are kept in high glucose–containing media for at least 1 week (37). The rate of glycation is so slow that rapidly turning-over intracellular proteins would not be present long enough to undergo nonenzymatic glycosylation (38). This is relevant to p53, which has a half-life of 20 min (10). The enzymatic O-glycosylation of p53 occurred between 10 and 45 min and progressively decreased after 3 h. The addition of O-GlcNAc to the COOH-terminal of p53 was replaced by phosphorylation, which became apparent only after 1 h of hyperglycemia. Phosphorylated p53 increased consistently from 3 h to 4 days. Inhibition of O-glycosylation prevented p53 phosphorylation at Ser 390, p38 MAPK activation, Ang II synthesis, and apoptosis. Thus, O-glycosylation and phosphorylation were sequentially implicated in the modulation of p53 function and transactivation of p53-dependent and -regulated genes in myocytes. Hyperglycemia-mediated O-glycosylation of p53 has never been shown before. Similarly, enzymatic glycosylation of p53 has never been documented in normal cells. The only example was obtained in a carcinoma-derived cell line in which wild-type glycosylated p53 is constitutively expressed and is biologically active (8). Glycosylated p53 promotes transcription of target genes and induces apoptosis.

Glycosylation and phosphorylation activate several transcription factors (7,8), ∼20 of which, including SP1, AP-1, SRF, and p53, are known to be O-glycosylated (8). The basic region at the COOH-terminus of p53 represses DNA binding, maintaining p53 in its latent form (9). This inhibition can be relieved by various mechanisms: introduction of bulky groups, including sugars like GlcNAc (9), or neutralization of charge through phosphorylation by PKC, casein kinase II, and p38-MAPK (9,10). The sites of glycosylation do not possess a common consensus sequence, but some shared characteristics exist (5). These sites are clustered at the COOH-terminus in the proximity of proline and valine residues. They are located in sequences rich in serine and threonine (5) that are similar to phosphorylation sites for protein kinases (4,5,6). These sites constitute the substrate for the catalytic activity of glycosyl transferase and protein kinases (6). The role of the hexosamine pathway in p53 glycosylation has been demonstrated here by the identification of O-GlcNAc residues bound to the transcription factor.

Phosphorylation of p53 was examined at three distinct sites: two at the COOH-terminus and one at the NH₂ terminus. In the first two cases, phosphorylation of the tumor suppressor was restricted to Ser 390 and was mediated by p38-MAPK. In contrast, there was no apparent p53 phosphorylation at Ser 376, indicating the lack of involvement of PKC in the activation of p53 with hyperglycemia. This conclusion was based on the reactivity detected with antibody 421, which recognizes p53 when phosphate groups (9,27,28,39) do not occupy the epitope 372–381. At the NH₂-terminus, p53 was phosphorylated at Ser 18, a site that is critical for p53 function in response to DNA damage (17). After DNA injury, phosphorylation at the NH₂ terminus of p53 is relevant to both the stabilization of this protein and its role in apoptosis and growth arrest (18). Cell death and block of the cell cycle in G₁ depend on p53 conformational changes associated with phosphorylation of Ser residues distributed at the NH₂ terminus (18). Ser 6, 9, 15, 20, and 46 are selectively phosphorylated by different kinases activated by DNA-damaging agents (16,17,18,40). DNA injury promotes a consistent phosphorylation of p53 at Ser 15 in humans and at its homolog in rodents, Ser 18 (17,18,41). We elected to evaluate phosphorylated p53 at Ser 18 for its critical role in the initiation of cell death as a result of DNA damage. Our findings on myocyte apoptosis and double DNA strand breaks with hyperglycemia are consistent with this possibility.

p38-MAPK, JNK, and ERK were investigated to identify their effects on hyperglycemia-mediated responses in myocytes. Downstream events triggered by ligand binding to AT₁ receptor were involved in phosphorylation of p38-MAPK (14). This signaling pathway was confirmed by the use of the AT₁ antagonist losartan. Treatment of myocytes with the AT₁ blocker had no influence on p53 O-glycosylation, but it abolished p53 phosphorylation. Losartan also prevented p38-MAPK activation. Inhibition of p38-MAPK by SB 202190 mimicked, at a more distal level, the consequences of losartan by preventing phosphorylated p53 and thus the synthesis and secretion of Ang II. The three MAPK families react differently to hormonal and metabolic stimuli, and the variability in their functional characteristics reflects the cell type involved (19,20,21). As shown here, hyperglycemia stimulates p38-MAPK and JNK in myocytes in a manner similar to vascular endothelial cells (19,21,42,43) and smooth muscle cells (44,45). Conversely, ERK is not activated by high glucose concentrations in these cell populations, but it is upregulated in adipocytes and fibroblasts (20). Ang II activates JNK through the AT₁ receptor in vascular smooth muscle cells and it inhibits ERK through the AT₂ receptor (45). Whether AT₂ receptors have a similar effect on myocytes is currently unknown. Because of the primary role played by p38-MAPK in myocytes exposed to hyperglycemia, we chose to selectively block this enzyme instead of using a more general inhibitor of MAPK, such as a MEK inhibitor. The latter would have affected the proapoptotic action of p38-MAPK and the antiapoptotic function of JNK (14,15,35,36), complicating the interpretation of the accumulated results.

Diabetes is characterized by upregulation of RAS (30,32), and interventions decreasing Ang II attenuate morbidity and mortality of diabetic patients (46,47). Inhibition of ACE positively modifies cardiovascular events and nephropathy in subjects with severe diabetes (47). However, the mechanism by which diabetes upregulates the systemic and/or local RAS remains to be identified. Ventricular myocytes possess a cellular RAS and synthesize and secrete Ang II (11). Additionally, Ang II stimulates cell death and cellular hypertrophy (11,24). The possibility that diabetes enhances the local RAS, resulting in the development of cardiac myopathy, has been shown in vivo (30,32); myocyte apoptosis and myocyte hypertrophy have been identified in the diabetic heart. AT₁ blockade prevents these phenomena as well as the induction of p53-regulated genes interfering with cell death and cell growth.

p53 has been linked to the transactivation of Aogen and AT₁ receptor, leading to the formation and release of Ang II from myocytes (11,12,22). p53 binding to the promoter of Aogen and AT₁ increased with high glucose concentrations, and this adaptation resulted in the accumulation of Aogen and AT₁ receptor proteins in myocytes. Aogen is the limiting factor in the generation of Ang II in these cells (22). Inhibition of p53 downregulates the expression of Aogen, attenuating the myocyte RAS and the synthesis of Ang II (12,22). In this study, glucose levels, p53 expression, and Ang II quantity have been found to be directly correlated. The increased Ang II in the medium 45 min after hyperglycemia most likely reflects de novo synthesis and release of Ang II from myocytes. This notion is consistent with the short half-life of the octapeptide, which is 6–12 min in the cells and 15 s in the blood (48,49). In the current experiments, the formation of Ang II followed the O-glycosylation of p53. Therefore, it is reasonable to assume that p53-mediated activation of the local RAS stimulated the generation and secretion of the peptide from the stressed myocytes.

Ligands binding to surface AT₁ receptors trigger myocyte death by a mechanism involving an increase in cytosolic Ca²⁺ and the activation of Ca²⁺-dependent DNase I (11,22). pH-dependent DNase II is not implicated in Ang II-mediated apoptosis (11,50). The present results are consistent with these observations, emphasizing the importance of hyperglycemia in enhancing the synthesis of Ang II and AT₁ receptor, which constitute the death signal and the executor of myocyte apoptosis, respectively. AT₁ and AT₂ receptors have distinct roles in the initiation of apoptosis in various cell types. In PC12W and R3T3 cells, which lack AT₁ receptors, Ang II stimulation of AT₂ results in internucleosomal DNA cleavage through dephosphorylation of the survival factor MAPK (51,52). In vascular smooth muscle cells, AT₂ receptor activation opposes ERK function, promoting apoptosis (53). AT₁ receptors induce cell survival by interfering with the effects of AT₂ on ERK activity (54). The claim has been made that neonatal and adult cardiac myocytes behave in a comparable manner in the presence of Ang II: AT₂ triggers apoptosis, and AT₁ inhibits cell death (55,56). The discrepancy between these results and our observations is difficult to explain. However, culture systems were not comparable, and much higher doses of Ang II, losartan, and PD123319 were used in these other experiments. These factors may account for the difference.

In conclusion, these in vitro results, in combination with in vivo observations (30,31,32), support the notion that hyperglycemia is responsible for apoptotic myocyte death with diabetes via activation of effector responses involving the local RAS. Importantly, a strong association between the upregulation of cellular RAS and the apoptosis of myocytes, endothelial cells, and fibroblasts has been identified in the diabetic human heart (31). Our work in diabetic patients (31) and experimental animals (30,32) is the only one that has emphasized the role of cell dropout in the initiation and progression of the diabetic myopathy. Because of the critical role of hyperglycemia in cell death, the current findings in vitro may apply to both juvenile type 1 diabetes and adult type 2 diabetes.

This work was supported by National Institutes of Health grants HL-38132, HL-39902, HL-43023, AG-15756, AG-17042, HL-66923, HL-65573, and HL-65577 and Juvenile Diabetes Foundation International grant 2000-62.

The initial helpful suggestions of Dr. Ashwani Malhotra are acknowledged.